The other feature that might be useful is the way to make async javascript function magically appear synchronous in python. As implemented, mpremote expects websocket input and reading from stdin to be blocking. So having some way (from python) to configure interpreter.sync would also be useful.

that's possible and it's already done in mpy terminal but it requires inevitably a worker executor to do so. This part is covered in our py-terminal plugin too ... when the script is a worker it uses sync readline or sync interceptor of chars

I have published micro-repl which uses PyScript and it's used in various projects already so maybe you can see in there how I deal with it in PyScript?

I've now looked at that repository, but I don't think it covers what I do. From what I see, it allows to use a micropython-pyscript in-browser, or to connect to a micropython device using "webserial", where no pyscript would be involved at all. This is not what I want to do. I want to run the existing mpremote utility (allowing file transfers, mounting, handling flow-control paste mode, mip, etc), inside a mpy-pyscript environment.

requires inevitably a worker executor to do so.

I'm fine with using a worker, this is already the part of py-terminal, I looked at (and where the initial Link I provided, points to). In the scenario where webrepl is currently run, it has full control over the headers, so activating the worker mode seems easy. Using a worker is not a problem for me, the problem is that configuring the sync object seems not possible from within the executed python code, but only if I adjust plugins/py-terminal/mpy.js which I would like to avoid.

This part is covered in our py-terminal plugin too ... when the script is a worker it uses sync readline or sync interceptor of chars

When the script uses a worker, there is a sync version of "input" (https://github.com/pyscript/pyscript/blob/main/pyscript.core/src/plugins/py-terminal/mpy.js#L144), that uses a minimalistic version of readline (https://github.com/pyscript/pyscript/blob/main/pyscript.core/src/plugins/py-terminal/mpy.js#L165), but I see not way to read individual characters (blocking) from python, and I don't see a way to configure the sync object, e.g. to add an additional sync.pyterminal_read_single_char from python.

the goal of this project is to make complicated things simple ... you are probably in the "moar complicated" "rabbit hole" and we have polyscript to address any underlying scenario where you can bootstrap your own thing and override the stdio as you wish. We don't have tons of documentation around our enabling bricks, as we don't prioritize these needs in here for the initial reason, but you can accomplish pretty much everything you want already although that'd be on the polyscript side of affairs ... and I hope you don't need to dig deeper in coincident or sabayon side of affairs, but the whole stack is there to make PyScript what it is and help developers needing more to dig into its stack.

This is to say: nothing is impossible, requirements and goals are different per each project in this stack though.

So the result is: both, reading single characters, and customizing sync hooks are not possible from within pyscript as-is but require cooperation from the surrounding website, did I get that correctly?

Because right now I see no way to access the interpreter or the sync objects in the main page.

the way we bootstrap the interpreter is though our own io, pyodide example here and that's way inside the stack: it's polyscript, not PyScript.

To interfere with that in PyScript you need to understand what our plugins do but if you want full control of the io in the interpreter, we don't have a story in here to enable that, but even if we did, it would be brought in here via polyscript.

Behind polyscript there's coincident and behind coincident there is sabayon ... and behind sabayon or coincident there is js-proxy ... this is to say, up to the polyscript project we deal with workers/main primitives, in polyscript we provide a library to fully take over anything you like about your interpreter.

In PyScript we enable with ease almost everything polyscript can provide with great control over what we need, or want, to provide but we also provide our own Python stdlib that still orchestrates what polyscript enables already.

I am not sure I am confusing you more than it's needed but you can see your very low level requirements are rather satisfied in polyscript project, not here, and if you manage to get what you want there, of course we'll welcome your suggestions in here too after understanding the requirements, the outcome, and the implementation.

Again, this is probably not what you wanted to hear but it's the current state of affairs and many others are (hopefully and apparently) happy with such current state. Too low-level control is a bit against this project goal as it exposes the project to more issues, not less, to deal with, for stuff that was likely not supposed to be "configurable".

I might cc @ntoll here just because he's way better than me in describing situations, and I'd be happy to learn from him, eventually, here there's some thing to change in our project ... I personally don't see it right now within this discussion.